Temperature and Density Dependence of the Viscosity of Toluene

octane, 0.5136c,d, 0.698 60, 16.342, 14.775, 823.5, 28.767 ... 0.3, 102.45, 0.899 37, 1.009, 268.0, 0.1, 102.47, 0.899 22, 0.9871, 280.2 ..... N = 1. ...
6 downloads 0 Views 62KB Size
J. Chem. Eng. Data 2000, 45, 893-897

893

Temperature and Density Dependence of the Viscosity of Toluene Kenneth R. Harris* School of Chemistry, University College, University of New South Wales, Australian Defence Force Academy, Canberra, ACT 2600, Australia

New measurements have been made for the viscosity of toluene between 255 K and 323 K at pressures up to approximately 400 MPa with a falling-body viscometer. These extend earlier high-pressure measurements below 298 K. The measurements form part of an intercomparison of high-pressure viscometers by the IUPAC Physical Chemistry Division, Commision 1.2, Subcommittee on Transport Properties. A revised correlation function is used to represent the results with a repeatability of (0.5%. The uncertainty is estimated at (1%.

Introduction New measurements have been made for the viscosity (η) of toluene between 255 K and 323 K at pressures up to approximately 400 MPa using a falling-body viscometer. These extend past high-pressure measurements below 298 K. The work forms part of a comparison of high-pressure viscometers of different types by the IUPAC Physical Chemistry Division, Commission 1.2, Subcommittee on Transport Properties. It is intended that toluene can become a standard reference substance for such instruments. Here we report only data obtained with our apparatus. They complement earlier measurements from this laboratory done at 298 and 323 K as a check on accuracy as part of a study on the viscosity of octane.1 In that paper, a correlation was produced using the best available data in the literature valid at pressures to 200 MPa. The equations employed reproduced individual data sets well but showed large residuals at high densities (where (dη/dF) is large), when all the available data sets were combined together. The present extension of the high-pressure falling-body data set to lower temperatures has been used to partially improve this situation. An account of the comparison of data sets obtained with falling-body, vibrating wire, torsionally oscillating crystal, and capilliary viscometers is to be published elsewhere.2 Experimental Section The toluene was High Purity Solvent grade from Burdick and Jackson (Muskegan, MI) (manufacturer’s analysis: >99.8% by GLC), dried over sodium wire. The density was determined, using matched pycnometers, to be (0.862 21 ( 0.000 01) g/mL at 25.000 °C. This compares well with the value obtained from the equation of state used in this and the previous work, 0.86223 g/mL, which is based on several sets of reliable experimental data.3 The molar mass was taken to be 92.1408 g/mol. The high-pressure viscometer and its operation have been described elsewhere.4 Temperatures below ambient were produced by cooling the thermostat fluid (Shell Diala B) with a heat exchanger coil through which cold alcohol from a Lauda UKW1500 refrigerated circulating unit was pumped. An on-off heater was controlled by a * To whom correspondence should be sent. E-mail: adfa.edu.au. Fax: 61 2 6268 8017.

k.harris@

bridge circuit employing a sensitive thermistor as one arm of the bridge to maintain temperatures constant to (0.005 K. The uncertainty of the primary Pt resistance thermometer (Tinsley Instruments, U.K.) employed is estimated at (0.01 K. The primary pressure gauge (400 MPa Heise CM, Dresser Instruments, Stratford, CT) was calibrated against a dead weight tester to (0.05%. The viscometer is automatic. New software using Microsoft Visual Basic replaced earlier Turbo-Pascal based software. Data are obtained in the following way. Under computer control, replicate (usually 3) measurements at a set temperature are made at atmospheric pressure. (The viscometer pressure vessel is inverted by a stepper motor to return the slug after each fall time is measured.) Then the viscometer is manually pumped up to a pressure a little above the maximum value required for the isotherm. Control is returned to the computer program, and the pressure is adjusted automatically to the highest of a set of pressures contained in a table in the computer program. This adjustment is accomplished by driving out the piston of a screw injector (model 37-5.75-60, High Pressure Equipment Co., Erie, PA) connected to the viscometer pressure vessel but lying outside the viscometer bath. A Wika model 891.01.2002 pressure transducer (Alexander Wiegand GmbH & Co., Klingenberg am Main, Germany) is used to monitor the pressure during this process. After a time interval sufficient for temperature and pressure equilibration, replicate (again usually 3) fall times are measured and the process is repeated through the series of set pressure points (see Table 2) until atmospheric pressure is reached. The calibration method was changed from that used previously which relied on an atmospheric pressure correlation for toluene over a range of temperatures recommended by Nieto de Castro and Dymond.5 Instead, a group of liquids was used for this purpose, with all calibrations being done at the single temperature, 25 °C. The liquids included a set of calibration standards obtained from the Cannon Instrument Co. (State College, PA) (Table 1) and samples of toluene, octane, and cyclohexane. The viscosity used for toluene was that used previously.1,5 The viscosities of the octane and cyclohexane samples were determined to (0.2% in an independently calibrated, flared-capillary glass Ubbelohde viscometer with a negligible kinetic energy correction.

10.1021/je000024l CCC: $19.00 © 2000 American Chemical Society Published on Web 08/16/2000

894

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000

Table 1. Calibration Data at 298.15 Ka substance (hexane)b

N4 octane toluene N8 (xylenes) cyclohexane N1 (mixed HCs) S3 (mineral oil) S6 (mineral oil) N10 (polybutene) S20 (mineral oil)

η/mPa s

F/g mL-1

t/s

t(1 - F/Fs)/s

Re

10-3A/Pa-1

0.3081 0.5136c,d 0.5542 0.6268 0.8980d 1.0030 3.4250 7.6800 15.1800 29.0000

0.663 60 0.698 60 0.862 19 0.861 60 0.773 89 0.789 60 0.841 10 0.856 90 0.865 10 0.853 90

9.816 16.342 18.043 20.445 28.850 32.313 111.170 249.520 493.530 940.620

8.922 14.775 15.908 18.027 25.785 28.811 98.335 220.170 434.923 830.367

2177 823.5 855.6 667.1 296.4 241.7 21.92 4.437 1.146 0.3106

28.958 28.767 28.704 28.760 28.740 28.725 28.711 28.668 28.651 28.633 28.707 ( 0.17%

average (disregarding N4)

a Sinker diameter ) 6.3 mm; density, F ) 7.285 g/mL. b N4, S1, and so forth are Cannon viscosity standard numbers. Calibration s viscosity values used are those supplied by Cannon for their materials. c The sample used is described in ref 1. The viscosity differs from d that recommended in ref 12. These values were determined with a glass Ubbelohde flared-capillary viscometer calibrated at 25 °C with water and aqueous sucrose solutions.13

Figure 1. Calibration plot: viscometer constant for various calibration fluids at 25 °C as a function of Reynolds number, Re.

The working equation for the falling-body viscometer is4,6,7

η(p,T) )

t(1 - F/Fs) A[(1 + 2R(T - Tref)][1 - 2β(p - pref)/3]

(1)

where t is the fall time, F is the density of the fluid, Fs is that of the sinker, R is the coefficient of thermal expansion (1.6 × 10-5 K-1) and β is the bulk compressibility (2 × 10-6 MPa-1) of the sinker and tube material, in this case 316 stainless steel. (The factor 2/3 in the second term of the denominator is incorrectly given as 2 in ref 4. The effect of this error is negligibly small.) A is the calibration constant. The sinker density was corrected for changes in T and p from the calibration state point, Tref ) 298.15 K and pref ) 0.1 MPa, using the relation8

Fs )

Fs(Tref,pref) [1 + 3R(T - Tref)][1 - β(p - pref)]

(2)

Some workers6 express A in terms of the quantity [t(1 - F/Fs)]N, where N is an arbitrary constant. For example, Malhotra et al.,4 calibrating with various fluids, used N ) 1. However in this work, A was found to be constant (N ) 0) within the experimental scatter (average deviation, (0.2%; maximum deviation, (0.3%). Figure 1 is a plot of A against the Reynolds number, Re. Equation 5 is not exact at high Reynolds numbers, and the point at which turbulence can occur is dependent on the value of the viscometer constant A. Inspection of graphs given by Isdale and Spence6 suggests the critical Re value is about 1000 to 2000 for A in the range 30 000 to 38 000 Pa-1, dropping to about 5 at A ∼ 3300 Pa-1. The value of A for the sinker used here was 28 707 Pa-1, which can be

expected to be valid for Re < 1000 at least. The point for hexane (the least viscous calibrant), which is somewhat high and for which Re ∼ 2180, was not included in this calibration average. For toluene, densities were obtained from an equation of state used previously.3 The uncertainty, based on comparisons with other high-pressure equations of state, is 93 cm3/mol), but there is a systematic upswing in the residuals below this molar volume where the terms ζ2Vr/ (1 + ζ3/Vr) and ζ1, which are of opposite sign, approach one another in magnitude. The unbiased fit of η (Figure 3) reduces the upswing and the scatter between isotherms at

Figure 2. Residuals for the indirect fit of the viscosity of toluene to eqs 3 and 4 obtained by fitting the Dymond reduced viscosity (η*) as a function of reference molar volume, Vr: ([) 255 K; (0) 258 K; (4) 268 K; (3) 278 K; (+) 288 K; (]) 298 K; (O) 323 K.

higher densities at the expense of the scatter at lower densities. Though the small range of the residuals is quite good, within (2%, the precision is clearly better, and the lack of randomness in the residuals shows that eqs 3 and 4 are not truly satisfactory in reproducing the data for this system. This was an unresolved question in our earlier study.1 Some effort has been expended to find an alternative set of equations to fit the data. Attempts were made to employ eq 3 with different forms of Vr: for example, (a) Vr ) V/V0,

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000 897

Figure 3. Residuals for the direct fit of the viscosity of toluene to eqs 3, 4, and 6 as a function of reference molar volume, Vr: symbols as in Figure 2.

Figure 5. Residuals for the direct fit of the viscosity of toluene to eqs 7 and 8 as a function of reference molar volume, Vr′: symbols as in Figure 2.

Acknowledgment I am grateful to other members of the IUPAC Division of Physical Chemistry, Commission 1.2, Subcommittee on Transport Properties, for helpful discussions. I particularly thank Prof. Marc Assael and Prof. Joa˜oFareleira as well as Prof. Akira Nagashima for his comment on the importance of checking Reynolds numbers when calibrating with low-viscosity fluids. Dr. Lawrie Woolf is thanked for criticism and advice. It is a pleasure to thank Mr. Shane Murtagh for assisting with the viscometer calibration and Messrs Kerry Richens and Ken Piper for electronics and mechanical workshop services, respectively. Figure 4. Plot of the quantity xT/η against molar volume, V. The isotherms shown, from top to bottom, are 323, 298, 278, and 258 K. Also shown is the curve linking the 0.1 MPa points for all the experimental isotherms.

V0 ) Lσ3/x2, and σ ) σ0 + σ1xT + σ2T; and (b) Vr ) V/V0 and V0 ) a + b(T - Tr) + c(T - Tr)2. It was possible to obtain satisfyingly small residuals and standard deviations with these forms for Vr, but the errors in the fitted coefficients were large and the values obtained varied considerably depending on the initial estimates made. In other words, the minima in the n-dimensional parameter surface were very shallow. Therefore, a variant of an equation successfully used with self-diffusion coefficients was tried. Figure 4 shows fluidity (φ ) 1/η) isotherms reduced by factoring out the primary kinetic theory 1/xT dependence. The isotherms are similar in the geometric sense, like those for the corresponding self-diffusion coefficient function D/xT.3,10 Also shown is the curve linking the 0.1 MPa points for all the experimental isotherms: this is nonlinear. (xT/η)/(xK/(mPa s)) ) 457.963 - 11.6496(V/(cm3mol-1)) + (7.16053 × 10-2)(V/cm3mol-1)2 with a standard deviation of (1.0%. For the complete data set, the best simple empirical function found was a Pade´ approximation combined with a simple mapping of the isotherms onto a single reference isotherm:

xT/η ) (ζ1 + ζ2Vr′ + ζ3Vr′2)/(1 + ζ4Vr′)

(7)

Vr′ ) V - ξ1(T - Tr) - ξ2(T - Tr)2

(8)

The coefficients are also listed in Table 3, and the residual plot is shown as Figure 5. The residuals cluster about the axis and are now better distributed (though not perfectly randomly) within a range of (1.3% and a standard deviation of (0.5%. Equations 7 and 8 therefore appear sufficient for the task in this case.

Literature Cited (1) Harris, K. R.; Malhotra, R.; Woolf, L. A. Temperature and Density Dependence of the Viscosity of Octane and Toluene. J. Chem. Eng. Data 1997, 42, 1254-1260. (2) Assael, M. J.; Avelino, H. M. T.; Dalaouti, N. K.; Fareleira, J. M. N. A.; Harris, K. R. Standard Reference Data for the Viscosity of Toluene over a Wide Range of Temperatures and Pressures. Int. J. Thermophys., submitted for publication. (3) Harris, K. R.; Alexander, J. J.; Goscinska, T.; Malhotra, R.; Woolf, L. A.; Dymond, J. H. Temperature and Density Dependences of the Self-diffusion Coefficients of Liquid n-Octane and Toluene. Mol. Phys. 1993, 78, 235-248. (4) Malhotra, R.; Price, W. E.; Woolf, L. A.; Easteal, A. J. Thermodynamic and Transport Properties of 1,2-Dichloroethane. Int. J. Thermophys. 1990, 11, 835-861. (5) Nieto de Castro, C. A.; Dymond, J. H. Private communication, Sept. 1996. (6) Isdale, J. D.; Spence, C. M. A Self-Centering Falling Body Viscometer for High Pressures. National Engineering Laboratory Report No. 592; Department of Industry, U.K., 1975. (7) Isdale, J. D.; Easteal, A. J.; Woolf, L. A. Shear Viscosity of Methanol and Methanol + Water Mixtures Under Pressure. Int. J. Thermophys. 1985, 6, 439-450. (8) Dymond, J. H.; Young, K. J.; Isdale, J. D. Transport Properties of Nonelectrolyte Liquid Mixtures. II. Viscosity Coefficients for n-Hexane + n-Hexadecane System at Temperatures from 25 to 100 °C at Pressures up to the Freezing Pressure or 500 MPa. Int. J. Thermophys. 1980, 1, 345-373. (9) Assael, M. J.; Dalaouti, N. K.; Dymond, J. H. The Viscosity of Toluene in the Temperature Range 210 to 370 K and at Pressures up to 30 MPa. Int. J. Thermophys. 1999, 20, 1367-1377. (10) Harris, K. R. Temperature and Density Dependence of the Selfdiffusion Coefficients of n-Hexane from 223 to 333 K and up to 400 MPa. J. Chem. Soc., Faraday Trans. 1 1982, 78, 2265-2275. (11) Dymond, J. H. The Interpretation of Transport Coefficients on the Basis of the van der Waals Model. I. Dense Fluids. Physica A 1974, 79, 100-114. (12) Dymond, J. H.; Øye, H. A. Viscosity of Selected Liquid n-Alkanes. J. Phys. Chem. Ref. Data 1994, 23, 41-53. (13) James, C. J.; Mulcahy, D. E.; Steel, B. J. J. Phys. D: Appl. Phys. 1984, 17, 225-230. Received for review January 24, 2000. Accepted June 7, 2000. This work was supported in part by grants from the Australian Research Council and the University College.

JE000024L